Solvent Effects in the Cyclization of Thiol Amides ... - ACS Publications

G. Alan Dafforn and Daniel E. Koshland, Jr.*. Contribution from the Department of Chemistry, Bowling Green State University,. Bowling Green, Ohio 4340...
1 downloads 0 Views 1MB Size
1246

Solvent Effects in the Cyclization of Thiol Amides and Hydroxy Acids G. Alan Dafforn and Daniel E. Koshland, Jr.* Contributionfrom the Department of Chemistry, Bowling Green State University, Bowling Green, Ohio 43403, and Department of Biochemistry, Uniuersity of California, Berkeley, Berkeley, California 94720. Received December 1.5, 1976

Abstract: Solvent effects on the ring closure of some thiol amides and hydroxy acids have been examined using sulfolane as a solvent. I n general, the relative reactivities in the dipolar aprotic solvent parallel those in water, but there are some deviations. Solvent effects on rates of the individual reactions studied here as well as on related reactions are generally modest, but examples of changes in rate as large as IO6 with changing solvent are noted if catalyst activity is considered. The role of solvation as a factor in reactions at an enzyme surface is strongly dependent on the specific reaction: solvation cannot be a universal factor in enzyme catalysis.

I. Introduction Among the effects which may contribute to the high catalytic efficiency of enzymes are possible solvation and microenvironmental effects. Both effects could presumably accelerate a chemical reaction by taking advantage of the sensitivity of many reactions to the nature of the surrounding medium. For example, a reaction which proceeds faster in a solvent of low dielectric constant might be accelerated by binding in a hydrophobic active site. More subtly, the structure of an enzyme active site might contain regions of very different solvating properties arranged in such a way as to desolvate the bound substrate but not the transition state for a reaction. Like many other effects contributing to enzymatic rates, this one is difficult to measure directly and is best approached through model systems. The simplest analogy to enzymatic solvation effects is the effect of a change in solvent on the rate of a model reaction such as ester hydrolysis.] It is widely appreciated that such solvent effects vary from moderate to many orders of magnitude, particularly when dipolar aprotic solvents are involved or a change in charge type occurs.2 Only limited efforts have been made to extrapolate these results to enzymatic reactions, leading to the conclusion that solvent effects certainly affect enzymatic reactions, but probably are not a dominant factor.' The question of microenvironmental effects is more difficult to deal with quantitatively. Although microheterogeneous environments can be created with micelles or polyelectrolyte^,^ the specificity of placement of charges, etc., in an enzymatic active site is difficult to mimic. Insofar as interpretation is possible, the generally modest catalysis by micelles (1 -2 orders of magnitude) would argue against large microenvironmental effects3 Related observations such as strong electrostatic catalysis by LiC104 in ether demonstrate the powerful catalytic effects of charge embedded in an apolar e n ~ i r o n m e n t Even .~ in this case, very high concentrations of salt are required to equal the solvating power of water. Thus the available evidence would seem to indicate a limited role for solvation and microenvironment effects; nevertheless, a number of reactions show quite spectacular solvent effects and the possibility of important solvation catalysis by enzymes cannot be dismissed. A related problem is the possible effect of solvation on the intramolecular reactions often used as models of enzymatic processes. The high rates of intramolecular reactions relative to bimolecular counterparts can be taken as evidence that placing two functional groups next to each other in a specific alignment, as at an enzyme active site, enhances their rate of reaction. In both enzymatic and intramolecular reactions, this effect could arise because the groups are desolvated relative Journal of the American Chemical Society

to each other. In both situations, the groups may be held so close that no solvent molecule can be interposed between them; compared to a bimolecular reaction, no solvent molecules must be removed from a solvation shell for reaction to occur. Such an effect would be of interest because observed rate enhancements of intramolecular over bimolecular reactions have been used as evidence for other theories of enzyme a ~ t i o nIn . ~particular, a large range of relative rates for acid-catalyzed lactonization or esterification in water of compounds IV-VI shown in Table I is observed.6 This and other examples of sensitivity of rate to structure have been interpreted in terms of orientation of functional groups. Although good evidence was presented that these relative rates were not a consequence of solvation, the possibility could not be totally ruled out. Structural and solvation effects on such a series of reactions could be completely separated by repeating the reactions in the gas phase or by gas-phase quantum mechanical calculations, but at least two kinds of problems arise in either case. Most reactions of interest as biochemical model systems do not occur in the gas phase, though such techniques as ion cyclotron resonance offer promise along these lines. Similarly, quantum mechanical calculations sophisticated enough to be believable remain extremely expensive and time consuming. In addition, the significance of such results is questionable since stripping away the solvent could substantially alter the mechanism.' A more feasible approach to the observation of solvation effects is to measure relative rates in two solvents of radically different character. This approach has been used by Bruice and co-workers8 to examine relative rates for intramolecular vs. bimolecular anhydride formation in water and MezSO-water. The individual reactions both involved nucleophilic attack by carboxylate on aryl esters and showed quite large solvent effects on going from water to 1 M H 2 0 in Me2SO. The ratio of the two rates was only slightly affected, indicating that solvation is probably not responsible for the high rate of the intramolecular reaction. We have undertaken a study of the effects of changing solvent on series of intramolecular reactions for two reasons. First, the extensive series of reactions used earlier as support for an orientational theory of high enzymatic rates were mostly acid catalyzed. Consequently, exclusion of solvation effects in our systems based on a single example of a reaction proceeding by a different mechanism was not proven. Second, the effect of large changes in solvation on rates of individual reactions might be useful as models for solvation effects in enzymatic reactions. The reactions chosen for reinvestigation in nonaqueous media are shown in Table I. Lactonization or esterification of the series IV-VI was an obvious choice as the hydroxy acids

/ 99.22 / October 26, 1977

1241

were the most thoroughly studied series earlier. Thiolactone formation from the series 1-111 was particularly interesting because of the unusual relative rates observed in water with I closing more rapidly than 11. Thus this series seemed a promising starting point for investigation of solvation. It also offered two potential experimental advantages: the thiolactone products absorb strongly in the UV region, making rates easy to measure, and the amide group might be completely protonated by an acid catalyst, simplifying the interpretation. Trifluoromethanesulfonic acid (HOTF) in sulfolane was chosen as the nonaqueous solvent system. The dipolar aprotic solvent sulfolane is quite similar to MezSO in most solvent properties such as dielectric constant and sufficiently different from water to expose any solvation effects. Sulfolane was preferable to Me2SO for our purposes because it apparently could be obtained transparent down to a t least 240 nm,9 the wavelength where thiolactones absorb,6 because it lacks the ability of MezSO to oxidize thiols,IOand because of its general high stability. Trifluoromethanesulfonic acid is an extremely strong acid]' but is unlikely to oxidize thiols. A strong acid catalyst offered the possibility of complete protonation of the amide starting materials, eliminating changes in pK, of the amides as a possible factor in relative rates. In addition, complications due to partial dissociation of the catalyst could be avoided with a sufficiently strong acid. 11. Experimental Section General. Kinetic measurements monitored at constant wavelength were made using a Gilford recording spectrophotometer Model 2000 with a cell compartment thermostated by a K-2/R Lauda/Brinkmann circulating bath or a Zeiss PMQ 11. UV spectra were determined using a Cary Model 14 with cell compartment thermostated a t 30 "C. Infrared spectra were taken with a Perkin-Elmer Model 257 grating infrared spectrometer using KBr pellets. N M R spectra were recorded on a Varian Associates A-60 spectrometer. Melting points were determined with a Thomas capillary melting point apparatus and were uncorrected. Elemental analysis and Karl-Fischer determinations were done by the Microchemical Analytical Laboratory, University of California, Berkeley. Lactones and Thiolactones. The preparation and purification of these compounds have been described previously.6 y-Mercaptobutyramide (I). y-Thiobutyrolactone was treated with anhydrous liquid ammonia in a Parr bomb for 24 h at room temperature. The product was contaminated by a significant amount of the disulfide dimer of y-mercaptobutyramide. The monomer was separated from the oxidation product by silica gel chromatography using acetone as a solvent. The product was stored under nitrogen in a desiccator to prevent oxidation of the mercapto group: mp 9 1-93 'C; I R (CHCI3) 1620, 1680 cm-' (amide I and I1 bands). Anal. (C4HgNOS) C , H , N , S. 2-endo-MercaptomethyIbicyclo[2.2.1]heptane-3-endo-amide (11). This amide was synthesized by treating 2-endo-mercaptomethylbicycl0[2.2.1] heptane-3-endo-carboxylic acid thiolactone with anhydrous liquid ammonia in a Parr bomb a t room temperature for 24 h. The product was chromatographed on silica gel using acetone as a solvent and recrystallized from benzene-n-hexane: mp 80-83 "C; IR (KBr pellet) 1680, 1620 cm-l (amide I and I1 bands). Anal. (C9HlsNOS) C, H, N, S. 2-endo-Carboxamido-6-endo-thiobicyclo[ 2.2.llheptane (111). The corresponding thiolactone (0.5-1 g) was placed in a 20-mL glass vial inside a thick-wall stainless steel Parr bomb. The bomb was capped with a drying tube and cooled to liquid N2 temperature. The vial was quickly filled with liquid NH3, capped, and allowed to stand at room temperature for 2 days. The bomb was then recooled with liquid N2, opened, and immediately placed in a lyophilizing flask. NH3 was removed by pumping overnight after careful evacuation to minimize bumping. The resulting fluffy white powder was transferred to vials under N2 in a dry bag and stored in the refrigerator. The amide was identified by IR (KBr pellet), showing N-H bands a t 3360 and 3190 cm-I and C=O bands a t 1700, 1660, and 1620 cm-l. Conversion was always incomplete. Extent of conversion was determined by the increase in UV absorbance at 240 nm on complete thiolactonization in HCl/HZO or sulfolane, and ranged from 10 to

75%. (More recent experience indicates that the product is most stable when stored at 0 OC over PzOs.) 2-endo-Hydroxymethylbicyclo[2.2.l]heptane-3-endo-carboxylic Acid (V). The lactone of the above acid was stirred overnight with a slight excess of I M N a O H and sufficient ethanol to solubilize lactone to give the sodium salt of the acid. The solution was cooled in an ice bath, and 1 equiv of acetic acid or concentrated HCI was added, giving an immediate precipitate. Extraction of the mixture with chloroform, separation, and evaporation of the chloroform layer gave the product as a white powder: I R 0 - H 3475, C=O 1705 cm-I (no 1760 cm-I lactone). 4-Hydroxybutanoic Acid (IV). This compound was too water soluble for isolation as described above. Instead, the free acid was obtained by potentiometric titration with methanolic HCI of sodium 4-hydroxybutanoate (obtained by hydrolysis of the lactone as above followed by lyophilization) in methanol. Addition of 5 volumes of ether and filtration removed NaCI, and the free acid was obtained as a yellowish oil by evaporation of solvent at low temperature: I R 3200-3450 (broad), 1720 cm-' (broad). Trifluoromethanesulfonicacid (HOTf)was obtained anhydrous from Minnesota Mining and Manufacturing Co. It was conveniently handled as an approximately 1 M stock solution in sulfolane, standardized by dilution in water and titration with standard N a O H solution. Sodium Trifluoromethanesulfonate. CF3S03H (1.91 g) was dissolved in about 20 mL of water, neutralized with dilute N a O H using a pH meter, and evaporated to dryness. The resulting white crystals were recrystallized from 5050 acetone-benzene, mp 251 S 2 5 2 . 5 OC (lit. mp 248 OC).lZ Ammonium Trifluoromethanesulfonate. CF3S03H (2.54 g) in 15 m L of water was neutralized to pH 6.6 with aqueous ammonia, evaporated to dryness, and recrystallized from 5 : 1 acetone-benzene, mp 225.5-226.5 "C. Hammett indicators were obtained as a set from Aldrich Chemical Co. All other compounds were of reagent grade. Sulfolane (Tetrahydrothiophene 1,l-Dioxide). Sulfolane was obtained from Phillips Petroleum Co. or from Aldrich Chemical Co. Aldrich material showed some visible light absorption and a solvent cutoff ( A = 1) a t 300 nm. Several purification procedures were tried, all based generally on the procedure used by Coetzee, Simon, and Bertozzi.13 The best of several procedures to remove material absorbing in the UV requires heating to 180-200 "C for 24 h under N2 in the presence of N a O H pellets, vacuum distillation of the resulting black liquid, stirring overnight with 30% fuming sulfuric acid, and redistillation. After standing over N a O H flakes for several days to remove sulfuric acid, the solvent was rapidly vacuum distilled in small batches (